合并返回 Pandas 中的 nan

Question

I want to merge df1 and df2 on common column ID . df2 looks like this:

    ID      TYPE   VALUE1 VALUE2    VALUE3
0   672117  Single 0.25   923.77    94.08
1   MSN242  DOUBLE 0.21   1219.31   105.77
2   673312  DOUBLE 0.20   4030.08   113.00
3   222255  Single 0.23   1119.38   126.69

code used:

df3 = pd.merge(df1, df2, on ='ID', how = 'left')

It seems there are 2510 common ID (all ID matched):

len(list(set(df1.ID) and set(df2.ID)))
>>> 2510

but df3 shows that all columns TYPE , VALUE1 , VALUE2 , VALUE3 are mostly nan . What went wrong?

---

Edit: df1 (shape 2510 rows × 22 columns ) looks like this:

    ID      CRITERION1  DATE                MEAS1       MEAS2       MEAS3    COMPOSITION    DPMT %CONTENT1  %CONTENT2 MeanGroup %Article1   %CA_Count   %CA_Count1  CATEGORY1 CATEGORY2 CODE    Group   COST1 COST2 COST3 COST4
0   000002  Y           2009-01-03 11:52:46 0.930150    17.412708   1.583333    Component   P   0.407859    0.979346    C   0.401572    0.000098    0.946168    Z   L   LEVEL1 NY   1767.0  1767.0  1767.0  1767.0
1   XC-004  Y           2009-01-03 11:52:46 1.898295    0.548192    0.250000    Component   NP  0.874263    0.999742    C   0.797250    0.000015    0.995345    Z   M   LEVEL1 NU   15525.0 15525.0 15525.0 15525.0

Answer 1

Since you merged (joined) left, it'll keep all IDs from the left table (df1) and drop all non-matching from df2. It then fills up all the non-existing VALUE1, VALUE2, VALUE3 from the IDs that are left-only with NaNs.

I'd assume your ID mismatch is pretty large and you have len(df1.ID) - 2510 number of NaNs rows in your table.